mance measures, but more importantly, it significantly facilitates the identification of corresponding failure modes and consequently the use of PHA and QI methods. Consistently (i.e., globally) defined errors across an entire health care setting enable a collective patient safety consciousness to address potential errors proactively, rather than retrospectively. Retrospective analysis, although necessary and insightful at times, is still retrospective—relying on performance measures of accidents/incidents or near misses, which themselves are the products of, or promote, hindsight bias and a host of other potential unwanted consequences.

Two studies illustrate the promises and shortcoming of CQI.

Concerns about the quality of health care in France led to the creation, in 1991, of a national agency for health care quality, Agence National pour le Développement de l’Evaluation Médicale (ANDEM). In 1997, ANDEM became Agence Nationale pour l’Accréditation et l’Evaluation en Santé (ANAES). Between 1995 and 1998, ANAES sought to increase hospital management’s awareness of CQI and to study its implementation in public hospitals. In 1995, a call was issued for projects on patient safety concerns such as nosocomial infections and incidents after anesthesia and blood transfusions. A second call for projects issued in 1996 was open for all project types. Selected projects received a financial incentive of between $10,000 and $80,000. Juries were composed of 12 to 14 individuals with experience in quality selected projects.

From 260 first-round project applications, 29 were selected and 26 were evaluated. Nine projects addressed prevention of nosocomial infections, five addressed medical records management, five addressed anesthesia safety, four addressed blood transfusion safety, four addressed drug dispensing safety, and two addressed controlling violence in psychiatric units. At evaluation, 38 months after initiation, 61 percent of the patient safety projects had met their objectives, and more than 50 percent of participating hospitals had established new CQI projects following the initial one. Half of the project team leaders considered that, at the time of the final evaluation, their main performance indicator (e.g., number of falls, number of nosocomial infections) had begun to evolve satisfactorily. Overall evaluation of this project is limited by the noncomparative nature of the study, which was